Temporal voice areas exist in autism spectrum disorder but are dysfunctional for voice identity recognition

Abstract

The ability to recognise the identity of others is a key requirement for successful communication. Brain regions that respond selectively to voices exist in humans from early infancy on. Currently, it is unclear whether dysfunction of these voice-sensitive regions can explain voice identity recognition impairments. Here, we used two independent functional magnetic resonance imaging studies to investigate voice processing in a population that has been reported to have no voice-sensitive regions: autism spectrum disorder (ASD). Our results refute the earlier report that individuals with ASD have no responses in voice-sensitive regions: Passive listening to vocal, compared to non-vocal, sounds elicited typical responses in voice-sensitive regions in the high-functioning ASD group and controls. In contrast, the ASD group had a dysfunction in voice-sensitive regions during voice identity but not speech recognition in the right posterior superior temporal sulcus/gyrus (STS/STG)-a region implicated in processing complex spectrotemporal voice features and unfamiliar voices. The right anterior STS/STG correlated with voice identity recognition performance in controls but not in the ASD group. The findings suggest that right STS/STG dysfunction is critical for explaining voice recognition impairments in high-functioning ASD and show that ASD is not characterised by a general lack of voice-sensitive responses.

Keywords: auditory; autism spectrum disorder; person identity recognition; superior temporal sulcus; voice recognition.

Fig. 1.

Experimental design of the two fMRI experiments. (A) In the vocal sound experiment, participants listened to blocks of vocal sounds (V), non-vocal sounds (NV), and silence (white boxes). One brain volume was acquired after each block. (B) In the voice identity recognition experiment, there were two conditions. In one condition, participants had to recognise who was speaking (voice identity task). In the other condition, participants had to recognise what was said (speech task). Stimuli consisted of blocks of 13 auditory sentences. At the beginning of each block, a key-word (‘Speaker’, ‘Speech’) on the screen instructed the participants to perform the voice identity or the speech task. Scans were acquired continuously. (C) Example trials of the voice identity recognition experiment: Participants decided for each sentence whether it was spoken by the target speaker (voice identity task) or whether it matched the content of the target sentence (speech task). Stimuli in the voice identity and speech task blocks were the same.

Fig. 2.

Behavioural results from the two fMRI experiments (see also Table 2 and Supplementary Table S3). (A) Performance accuracy in the voice identity recognition experiment: The ASD group performed significantly worse than the control group in the voice identity task. There were no significant differences between the ASD and the control group in the speech task. (B) Total amount of recalled sounds after the vocal sound experiment. The ASD group recalled significantly less non-vocal and vocal sounds than the control group. The vocal sound condition contained both speech and non-speech sounds. The ASD group recalled a comparable number of speech sounds but less non-speech sounds as compared to the control group. Error bars represent± 1 SE; *P < 0.05; **P < 0.005; n.s. not significant.

Fig. 3.

Vocal sound experiment. (A) Contrast vocal sounds > silence baseline. The control group as well as the ASD group showed BOLD responses along the STS/STG when listening to vocal sounds. The figure displays results for the right STS/STG, for the left STS/STG see Supplementary Table S2. (B) Contrast vocal > non-vocal sounds. Both groups showed enhanced BOLD responses in the right STS/STG when listening to vocal, compared to non-vocal, sounds. The results are displayed at the threshold of P = 0.05 FWE corrected for the right STS/STG. They are overlaid onto a group specific average of normalised T1- weighted structural images. Colour bars represent t-values.

Fig. 4.

Voice identity recognition experiment. (A) Contrast voice identity task > silence baseline. The control group as well as the ASD group showed BOLD responses along the right STS/STG when the task was to recognise voice identity. (B) Contrast voice identity task > speech task. The control group showed greater BOLD responses when recognising voice identity compared to when recognising speech. In the right posterior STS/STG these responses were higher for the control group as compared to the ASD group. (C) In the control, but not in the ASD group, responses in the right STS/STG to voice identity recognition correlated positively with performance in voice identity recognition. This correlation was stronger in the anterior STS/STG in the control group as compared to the ASD group. Results are presented for the right STS/STG and overlaid onto a group specific average image of normalised T1- weighted structural images. The results are significant at P = 0.05 FWE corrected for the ROI. For display purposes only the threshold of P = 0.01 uncorrected was used. Colour bars represent t-values.